ANALYSIS OF SEGMENTATION PARAMETERS IN ECOGNITION SOFTWARE USING HIGH RESOLUTION QUICKBIRD MS IMAGERY

Abstract

For object- oriented classification approaches, main step is the segmentation part of the imagery. In eCognition v.4.0.6 software, with the segmentation process, meaningful objects can be created for following steps. In the experimental imagery with 2.4m ground sampling distance (GSD) has been used and several different parameters e.g. scale, color/shape and smoothness/compactness parameters have been tested accordingly. Additionally, segmentation parameters were set to low and high values and thus, dissimilarity of segmentation results were examined.